Actually, I suspect that the X-Box will not ship with OpenGL drivers, just to make sure X-Box API is incompatible with an open standard. While NVidia can (and will) write OpenGL driver for the PC version of the card, I'd be surprised if Microsoft included it in the X-Box. This is the tactic they used ever since Windows NT 3.1, when they could have included NFS support but refused to. (perhaps there's even earlier example, but that's the first one that comes to mind). Heck, just recently MS desided to remove Netware support from the next release of Windows, but reconsidered due to customer outrage. ___
First of all, this API is crap. Anyone who has done some coding with OpenGL wouldn't touch DirectX with a 5 foot long stick. Second (related to first) DirectX is too tightly tied to COM and win32. (Surprised? Why would M$ implement a generic API?) Implementing DirectX would essentially involve implementing COM and parts of win32. There are people who've been doing this for quite a while. Unsuccessfully. ___
Corel didn't really port WordPerfect Office. They just got it to run under WINE. As such, it certainly shouldn't be called "WordPerfect Office for Linux". Is this also a WINE "port" or a real port?
Sorry, no benchmarks exist to prove it one way or the other. Most of these claims come from either *BSD or Linux zealots, so they are not very credible.
I did run FreeBSD 4.0 for a few days. I did notice some speed increase, but I wouldn't call it "so much faster". I noticed that it is significantly more efficient when it comes to swapping. Perhaps it's due to a better implementation of virtual memory. Anyway, it swapped quite a bit less than Linux running KDE with 64Mb RAM. But virtual memory is being reworked in 2.4, so we'll see some major improvement. However, the BSD filesystem was much slower than Linux's. It took forever to delete a directory with lots of files in it. I tried the same thing in Linux and it was several times faster. Also, Linux currently scales better than *BSD. SMP support in *BSD is somewhat like it was in 2.0 kernel. And again, 2.4 has some major SMP improvements. Overall though, I would say that FreeBSD was indeed marginally faster than Linux on my machine. But this is once again nothing more than anecdotal evidence. ___
Uhhm, no. If you use Outlook, it is enough to open the message. Also, you can't just double click on a java class to run it (even on windows). You need to save it to a directory, and then type in the command with the right classpath and the class name. ___
And what's the point of that? You cannot run this code as an applet, so you cannot embed it in an email and send it to people. You can with VBscript, ActiveX, and from the looks of things C-hash as well.
This code above will work *only* if you compile it to a.class. But then you need to explicitely run it yourself using the JVM. So you cannot write the Melissa/ILOVEYOU-style worms in Java. ___
BTW. latest Linux kernel does include httpd deamon... so even if MS did put parts of IIS in the kernel why the hell Linus and others do the same if it is such a ugly feature ?
That's what I was wondering too. My jaw fell off when I first saw this "feature" on the list of changes. Why did Linux devevlopers feel compelled to copy Microsoft by trading reliability for a little speed increase? The only consolation is that you can un-compile khttpd, so that you don't have a web server sitting in your kernel. Can't do it with IIS though, so NT boxes are always stuck with a web server in the kernel. ___
I was under impression that Java doesn't need templates because every object is automatically derived from the Object class. So to implement a generic container class, all you need to do is implement it for Object. Correct me if I'm wrong.
And how is this different from a C binary (exe) or a Java class (class). The both have full access to the OS don't they? Yes they do. I can write a java program to delete *.mp3 just as easily as I can write a COM object to do the same thing. Yeesh.
You are either a troll or are really that ignorant. Java applets do not have full access to the OS. In fact they do not have access to the file system period. You can read/write files only if you compile the Java code as an application, but than you can't run it in a browser. Oh, and since you have no pointers at all, you never have direct access to memory either.
First of all IIS does run in the kernel (at least part of it). Second, users didn't have any say over what to put in the kernel or not. Perhaps the reliability hit is acceptable for a workstation, but please explain me why would you want to run GUI in the kernel on a *server*? Or, even better, why would you want to run GUI at all?
As for no one wants a ported version of Windows 2000, that's only part of the story. NT 4 used to run on MIPS, Alpha and PowerPC. SGI ported it to MIPS, Digital ported it to Alpha, and I don't remember who ported it to PowerPC. But anyway, Microsoft had no intention of porting it. Digital (now Compaq) was the last company to say "we will no longer port your OS to our hardware for you" and Microsoft immediately killed all future plans for NT on Alpha, and made it out as though it was Compaq's fault. ___
But don't forget that the main guy behind the original NT architecture was Dave Cutler, a VAX guy. According to the original spec, NT was supposed to be a solid system with no GUI in the kernel, no "DirectX" crap in the kernel, no IIS in the kernel. It was also supposed to be microkernel-based with enough hardware abstraction to make porting easy. etc.
Look where NT is now. Bloated, buggy, unstable, insecure, and as of Windoze 2000 it runs on only 1 platform - x86. So much for the spec. You just can't make a good product when markedroids are involved in the design.
You are obviously completely ignorant about the issue. The whole point is that Rambus did *not* developed anything. There's prior art for every patent Rambus holds, which makes the patents invalid. And Rambus *knows* that if it the matter goes to court, they will have no claim. That is why they haven't gone after Micron or any other big memory players.
The really crappy thing is that Toshiba decided to settle without a fight. But that's not too bad yet. So far no US company decided to cave, and Micron has already said they'll fight. The Toshiba's "licencing" will have no relevance if/when they do take it to court. Also, only one court case will be enough to completely ivalidate Rambus's "patents".
Anyway, next time check your facts instead pulling nonsence out of your ass. ___
The difference is that the morons at Sharkey Extreme were trying to overclock the Duron by raising the bus speed. But wonder of wonders! AMD chips are not crippled by the multiplier lock like the Intel ones (do the sharkey guys even know that?). Tom actually got the Asus board that allows you to switch the multiplier and core voltage. So you can set the 650MHz chip (6.5x100) to 950MHz (9.5x100) by changing the multiplier. Can't do that to Intel CPUs.
I find that in general, Sharkey is pretty low on content. I'd trust Tom over Sharkey any time.
The only thing Sun machines have that commodity x86 PCs don't (well, besides the label...) is the 64 bit architechture. That is actually very important for the big-ass database servers that have several gigabytes of RAM. The 32 bit architecture is limited to only 4 GB of RAM, which is not enough for large-scale DB servers. But 64-bit or 32-bit is irrelevant for a workstation that only has like 128-256MB of RAM.
Oh, and there's the CPU scalability as well. SPARC architecture scales up to 64 CPUs. Intel boxes can just barely scale up to 4 CPUs, and even than, from what I heard, not all Xeons actually work properly in 4x configuration.
So, Sun boxes are good for the high end. However, as you correctly noticed, on the low-mid range PCs running Linux provide the same or better performance at much lower prices. But the low-end Sun boxes are expensive for the same reason Sony is expensinve. ___
Great points, a42. I too feel the same way. Java has its place and purpose but there is simply no way in hell it will replace C++. Much less in 5 years, as the troll above suggested. There are some things I like about Java (such as pure-OO, garbage collection, built-in support for exceptions and threads, extensive library, etc.) and some things I dislike (such as lack of MI, garbage collection (again), bloat, and speed (or lack thereof), etc.)
Oh, and speaking of garbage collection, I do believe it can be very useful if done right. It's just that the way it's implemented in Java totally blows. I think explicit delete statement is a must (you already gave a great example why). Garbage collection should augment, but not impede manual memory management. (i.e. it's nice to have the VM delete stuff for me if I *forget* to delete something, but the ability to explicitly delete stuff is a must, IMHO). Also, all things considered, smart pointers are better than the fancy memory-sweeping GC. Sure they don't work for circular linked lists and such, but 1) How often do you have to deal with those? 2) If you are writing a class that implements a circular linked list or a graph or whatever, you'd better know what you're doing! On the plus side, smart pointers have almost no overhead and are very easy to implement. ___
being logged in as root is actually not a good idea. I will not run any proprietary binaries as root, particularly something as big and bloated as star office. Instead I created a special (unpriviliged) user account called install. Before running the installer, I created the/opt/StarOffice directory as root, and chown'ed it to install. Then I ran the installer as install. This way, the only place the SO installer can write to is/opt/StarOffice, which guarantees that it will not fuck up my system. After the install is done, its a simple matter of running "chown -R root:root/opt/StarOffice" to fix the file ownership.
I don't get it. Why would you *want* to play 9 files at a time? You can't possibly listen to more than one at a time. I can only imagine what the noise was like. ___
First of all, this does not look like LinuxOne. Or, I should say, it looks less vapourous than LinuxOne. Here we have a company that actually knows what it wants to produce. Whether or not it's going to succede is a different matter altogether. LinuxOne, on the other hand, was founded by a spammer, had no product and no business plan, and, to top it off, filed for IPO within the first month of its existance.
Secondly, your statement "everyone around here try so dang hard not to say a bad word about it" is false. Most posts I see are pointing out problems with this console. And I agree that this thing is even less likely to succede than Micro$oft's X-Box.
so, if I install Motif on Linux instead of Lesstif and run DDD, I will be violating GPL, right? Also, does that mean that Sun, for example, cannot ship DDD with Solaris even if it wanted to? For that matter, Sun wouldn't be able to legally ship any of the GNU tools.
You need to read the GPL license again, dude. You can't link to non-GPL compatible software unless it is part of the OS. Hence, GPL Motif applications are fine on Solaris but not on Linux.
a-ha! So why don't you make QT a part of the OS? QT even meets Debian's free software guidelines, as you stated before. Oh, and BTW, does this mean that ddd (www.gnu.org/software/ddd/) is illegal to run under Linux? ___
Slashdot Mirror
This article *was* posted on slashdot. Last year, when it was news. Now it's history. Next time look in the older articles before claiming nonsense.
moderators: this is a troll. Moderate accordingly.
___
Actually, I suspect that the X-Box will not ship with OpenGL drivers, just to make sure X-Box API is incompatible with an open standard. While NVidia can (and will) write OpenGL driver for the PC version of the card, I'd be surprised if Microsoft included it in the X-Box. This is the tactic they used ever since Windows NT 3.1, when they could have included NFS support but refused to. (perhaps there's even earlier example, but that's the first one that comes to mind). Heck, just recently MS desided to remove Netware support from the next release of Windows, but reconsidered due to customer outrage.
___
First of all, this API is crap. Anyone who has done some coding with OpenGL wouldn't touch DirectX with a 5 foot long stick.
Second (related to first) DirectX is too tightly tied to COM and win32. (Surprised? Why would M$ implement a generic API?) Implementing DirectX would essentially involve implementing COM and parts of win32. There are people who've been doing this for quite a while. Unsuccessfully.
___
any idea when Debian will be released? Will it be this year/century? I'm tired of waiting for it already.
___
Corel didn't really port WordPerfect Office. They just got it to run under WINE. As such, it certainly shouldn't be called "WordPerfect Office for Linux". Is this also a WINE "port" or a real port?
___
Sorry, no benchmarks exist to prove it one way or the other. Most of these claims come from either *BSD or Linux zealots, so they are not very credible.
I did run FreeBSD 4.0 for a few days. I did notice some speed increase, but I wouldn't call it "so much faster". I noticed that it is significantly more efficient when it comes to swapping. Perhaps it's due to a better implementation of virtual memory. Anyway, it swapped quite a bit less than Linux running KDE with 64Mb RAM. But virtual memory is being reworked in 2.4, so we'll see some major improvement. However, the BSD filesystem was much slower than Linux's. It took forever to delete a directory with lots of files in it. I tried the same thing in Linux and it was several times faster. Also, Linux currently scales better than *BSD. SMP support in *BSD is somewhat like it was in 2.0 kernel. And again, 2.4 has some major SMP improvements. Overall though, I would say that FreeBSD was indeed marginally faster than Linux on my machine. But this is once again nothing more than anecdotal evidence.
___
Uhhm, no. If you use Outlook, it is enough to open the message.
Also, you can't just double click on a java class to run it (even on windows). You need to save it to a directory, and then type in the command with the right classpath and the class name.
___
And what's the point of that? You cannot run this code as an applet, so you cannot embed it in an email and send it to people. You can with VBscript, ActiveX, and from the looks of things C-hash as well.
.class. But then you need to explicitely run it yourself using the JVM. So you cannot write the Melissa/ILOVEYOU-style worms in Java.
This code above will work *only* if you compile it to a
___
That's what I was wondering too. My jaw fell off when I first saw this "feature" on the list of changes. Why did Linux devevlopers feel compelled to copy Microsoft by trading reliability for a little speed increase? The only consolation is that you can un-compile khttpd, so that you don't have a web server sitting in your kernel. Can't do it with IIS though, so NT boxes are always stuck with a web server in the kernel.
___
Hi Lars. Done question 3a yet? ;-)
___
I was under impression that Java doesn't need templates because every object is automatically derived from the Object class. So to implement a generic container class, all you need to do is implement it for Object. Correct me if I'm wrong.
___
You are either a troll or are really that ignorant. Java applets do not have full access to the OS. In fact they do not have access to the file system period. You can read/write files only if you compile the Java code as an application, but than you can't run it in a browser. Oh, and since you have no pointers at all, you never have direct access to memory either.
___
First of all IIS does run in the kernel (at least part of it). Second, users didn't have any say over what to put in the kernel or not. Perhaps the reliability hit is acceptable for a workstation, but please explain me why would you want to run GUI in the kernel on a *server*? Or, even better, why would you want to run GUI at all?
As for no one wants a ported version of Windows 2000, that's only part of the story. NT 4 used to run on MIPS, Alpha and PowerPC. SGI ported it to MIPS, Digital ported it to Alpha, and I don't remember who ported it to PowerPC. But anyway, Microsoft had no intention of porting it. Digital (now Compaq) was the last company to say "we will no longer port your OS to our hardware for you" and Microsoft immediately killed all future plans for NT on Alpha, and made it out as though it was Compaq's fault.
___
But don't forget that the main guy behind the original NT architecture was Dave Cutler, a VAX guy. According to the original spec, NT was supposed to be a solid system with no GUI in the kernel, no "DirectX" crap in the kernel, no IIS in the kernel. It was also supposed to be microkernel-based with enough hardware abstraction to make porting easy. etc.
Look where NT is now. Bloated, buggy, unstable, insecure, and as of Windoze 2000 it runs on only 1 platform - x86. So much for the spec. You just can't make a good product when markedroids are involved in the design.
___
You are obviously completely ignorant about the issue. The whole point is that Rambus did *not* developed anything. There's prior art for every patent Rambus holds, which makes the patents invalid. And Rambus *knows* that if it the matter goes to court, they will have no claim. That is why they haven't gone after Micron or any other big memory players.
The really crappy thing is that Toshiba decided to settle without a fight. But that's not too bad yet. So far no US company decided to cave, and Micron has already said they'll fight. The Toshiba's "licencing" will have no relevance if/when they do take it to court. Also, only one court case will be enough to completely ivalidate Rambus's "patents".
Anyway, next time check your facts instead pulling nonsence out of your ass.
___
The difference is that the morons at Sharkey Extreme were trying to overclock the Duron by raising the bus speed. But wonder of wonders! AMD chips are not crippled by the multiplier lock like the Intel ones (do the sharkey guys even know that?). Tom actually got the Asus board that allows you to switch the multiplier and core voltage. So you can set the 650MHz chip (6.5x100) to 950MHz (9.5x100) by changing the multiplier. Can't do that to Intel CPUs.
I find that in general, Sharkey is pretty low on content. I'd trust Tom over Sharkey any time.
___
The only thing Sun machines have that commodity x86 PCs don't (well, besides the label...) is the 64 bit architechture. That is actually very important for the big-ass database servers that have several gigabytes of RAM. The 32 bit architecture is limited to only 4 GB of RAM, which is not enough for large-scale DB servers. But 64-bit or 32-bit is irrelevant for a workstation that only has like 128-256MB of RAM.
Oh, and there's the CPU scalability as well. SPARC architecture scales up to 64 CPUs. Intel boxes can just barely scale up to 4 CPUs, and even than, from what I heard, not all Xeons actually work properly in 4x configuration.
So, Sun boxes are good for the high end. However, as you correctly noticed, on the low-mid range PCs running Linux provide the same or better performance at much lower prices. But the low-end Sun boxes are expensive for the same reason Sony is expensinve.
___
Great points, a42. I too feel the same way. Java has its place and purpose but there is simply no way in hell it will replace C++. Much less in 5 years, as the troll above suggested. There are some things I like about Java (such as pure-OO, garbage collection, built-in support for exceptions and threads, extensive library, etc.) and some things I dislike (such as lack of MI, garbage collection (again), bloat, and speed (or lack thereof), etc.)
Oh, and speaking of garbage collection, I do believe it can be very useful if done right. It's just that the way it's implemented in Java totally blows. I think explicit delete statement is a must (you already gave a great example why). Garbage collection should augment, but not impede manual memory management. (i.e. it's nice to have the VM delete stuff for me if I *forget* to delete something, but the ability to explicitly delete stuff is a must, IMHO). Also, all things considered, smart pointers are better than the fancy memory-sweeping GC. Sure they don't work for circular linked lists and such, but 1) How often do you have to deal with those? 2) If you are writing a class that implements a circular linked list or a graph or whatever, you'd better know what you're doing! On the plus side, smart pointers have almost no overhead and are very easy to implement.
___
Is Corel ever gonna start *porting* their apps to Linux? WINE just doesn't cut it for me.
___
being logged in as root is actually not a good idea. I will not run any proprietary binaries as root, particularly something as big and bloated as star office. Instead I created a special (unpriviliged) user account called install. Before running the installer, I created the /opt/StarOffice directory as root, and chown'ed it to install. Then I ran the installer as install. This way, the only place the SO installer can write to is /opt/StarOffice, which guarantees that it will not fuck up my system. After the install is done, its a simple matter of running "chown -R root:root /opt/StarOffice" to fix the file ownership.
___
I don't get it. Why would you *want* to play 9 files at a time? You can't possibly listen to more than one at a time. I can only imagine what the noise was like.
___
First of all, this does not look like LinuxOne. Or, I should say, it looks less vapourous than LinuxOne. Here we have a company that actually knows what it wants to produce. Whether or not it's going to succede is a different matter altogether. LinuxOne, on the other hand, was founded by a spammer, had no product and no business plan, and, to top it off, filed for IPO within the first month of its existance.
Secondly, your statement "everyone around here try so dang hard not to say a bad word about it" is false. Most posts I see are pointing out problems with this console. And I agree that this thing is even less likely to succede than Micro$oft's X-Box.
So, it sure looks like you're just trolling.
___
so, if I install Motif on Linux instead of Lesstif and run DDD, I will be violating GPL, right? Also, does that mean that Sun, for example, cannot ship DDD with Solaris even if it wanted to? For that matter, Sun wouldn't be able to legally ship any of the GNU tools.
___
a-ha! So why don't you make QT a part of the OS? QT even meets Debian's free software guidelines, as you stated before. Oh, and BTW, does this mean that ddd (www.gnu.org/software/ddd/) is illegal to run under Linux?
___
Oh, and it's xmms, BTW (X multimedia system). There is no such thing as xmps.
___